thread max_stack() returns always aligned stacksize-4 in rtos5

[kseverinkangas-zg]

Description

• Type: Bug
• Priority: Minor

---

Bug

Target
K64F

Toolchain:
ARM

Toolchain version:
5.06
mbed-cli version:
1.1.1

mbed-os sha:
78474a5

Expected behavior
Should return maximum stack memory usage to date

Actual behavior
Calling max_stack() returns: aligned stacksize-4
Steps to reproduce

---

Enhancement

Reason to enhance or problem with existing solution

Suggested enhancement
thread stack should be filled with the osRtxStackFillPattern pattern instead of hardcoded 0xE25A2EA5U.
rtx_thread.c already sets the first byte to osRtxStackMagicWord (0xE25A2EA5U) and the rest is filled with the osRtxStackFillPattern.

[0xc0170]

This is the code for max stack in Thread.cpp (https://github.com/ARMmbed/mbed-os/blob/master/rtos/Thread.cpp#L292).

I do not recall why are are not using by default the osThreadGetStackSize and osThreadGetStackSpace. I assume as we fill the stack with our own defined value (Thread::start method does it) therefore this is used instead. If these functions are used instead, does it return the expected value?

If stack stats are enabled, then OS_STACK_WATERMARK is set to 1 , and RTX watermark is used.

cc @c1728p9

[c1728p9]

The guard word might be causing max_stack to be returning the wrong value.

@0xc0170 if osThreadGetStackSize and osThreadGetStackSpace are used it should fix the problem, as long as watermarking is always on. Here is the previous reason why this was left out:
#4776 (comment)

[ciarmcom]

ARM Internal Ref: MBOTRIAGE-796

[kegilbert]

@c1728p9 Picked this up from the TRIAGE bot, are the comments you linked to still valid/what should be done with this issue?

[SenRamakri]

Fix in progress - #7545

[SenRamakri]

PR #7545 has been merged.